How a simulation tool can dramatically improve warehouse operations.

The key technology involved here may be providential for many warehouse managers: a simulation engine that evaluates product layouts versus slotting and picking strategies, the outcomes of which validate the best approach using actual cost estimates and critical metrics for each option.

A simulation engine is a software program that can replicate material movements in a virtual warehouse environment. It is based on geometry, on-hand inventory, material supply and demand. To be able to generate convincing simulations, three essential elements are required:

• Warehouse modeler, for geometrical information and locations attributes
• SKU inventory, they populate specific locations in the virtual warehouse
• Order profiles, existing or forecasted they drive warehouse operations

The simulation engine is demand-driven: for example, an order profile establishes the demand, and from the sequenced order lines, picking tasks are created, and then products are pulled out locations. According to rules of supply (replenishment), and constraints from the selected slotting configuration options, the virtual stocks are replenished when required. As the simulation engine clock goes forward, the warehouse operations are simulated, from material receipts to stock shipments, complete with travel and material handling.
Picking is done along ordered sequences, fixed paths, or optimized paths. SKU types and location attributes determine the pick type and associated cost (forklift, carton pick, etc). The number of item lines to be picked during one run may vary (single pick, batch picking, by order, by zone, etc).

While the simulation engine is running, operation costs and various metrics are calculated and logged. There’s a cost associated with every moves: traveling to and from pick locations per transport type, picking time per method, replenishment travel from and to reserve area. Mobile equipment (human, forklift, etc) have cost attributes that are factored in the calculations. Many assumptions and heuristics must be used to achieve a valid simulation of warehouse operations. Yet, the simulation should not be about literally replicating human activities in a warehouse as such, but should be focused on material movement and its cost.

The simulation engine does not dictate what to do. Key performance indicators are used to evaluate the relative outcomes of different scenarios. Thus, different configurations can be compared with each other. The best solution will stand out clearly if it exists, and alternatives might reveal how significant or not are the benefits of a specific strategy.

The ultimate goal is to minimize cost, but more specifically it’s to optimize pick efficiency, shipping performance, space utilization and safety. When comparing the efficiency of warehouse operations using different strategy simulations, we look for improvement in any one or more of key performance indicators (KPI).

Advanced Features

There is a higher calling for the simulation engine: to provide a playing ground for testing ideas and trying alternate strategies.

A Warehouse Modeler gives flexibility and modularity option for the simulation engine. Different parts of the warehouse may have different supply/demand behaviors and may require different slotting strategies. This capacity can be used to investigate hypothetical layouts like new cross-aisle bridges, restructured zone layout and storage modes distribution, or new racking requirement. Hypothetical products and sales forecast can be used as inputs, to test the warehouse capacity to adapt.

Multiple simulation engine instances can be daisy chained to represent a broader view of the supply chain. Here are a few interesting applications of this advanced feature.

• Reserve and forward pick areas configurations
• Distributed Optimization: connecting distribution centers to retail centers
• The manager of a DC may design slotting and picking strategies, not to optimize local operations, but to maximize efficiency for the retail center down the road, which may have a higher put-away and picking cost.

Limitations and benefits

Simulation engine performance is limited by available computing time and memory. Order profiling and forecasting are very sensitive to data corruption, variations and exceptions.
In any case, simulations are to be evaluated using judgment and experience. But the opportunity of testing many different hypotheses, however odd these may be, comes without any additional cost in this virtual world.
Used wisely, simulations provide a great decision tool.